Invisible / Unicode Prompt Injection

← AI/ML (OWASP LLM/GenAI Top 10)

unicode_invisible_prompt_injection   MEDIUM

Description

Hidden instructions encoded with zero-width / RTL-override characters in scanned or shared text are obeyed by the assistant.

How it works

A shared note looks benign but hides an instruction using zero-width characters and an RTL override. The app’s sanitizer only touches ASCII whitespace, so the invisible instruction survives and the assistant obeys it. A correct sanitizer strips all zero-width and formatting control characters before trusting the text.

How to exercise it. DVMA is the harness - open this module from the home index and tap the demo action. The screen ships the malicious input and simulates the attacker (e.g. the companion app, crafted intent, or scanned payload) in-process, and the evidence panel prints the proof. The Tools (optional) and Attack inputs below are only needed to reproduce the exploit end-to-end on a real device.

Exploit steps

1. Set up. Build DVMA with a flavor that enables the AI/ML (OWASP LLM/GenAI Top 10) category (e.g. --dart-define-from-file=config/flavors/dev.json) and run on an emulator/simulator you control. The demo needs no external tooling; for the optional on-device reproduction the relevant tools are: garak, promptfoo.
2. Locate the target. From the home index, open Invisible / Unicode Prompt Injection (unicode_invisible_prompt_injection). The How it works section above describes this module’s specific weakness; the screen states the intended-secure behavior and exposes the vulnerable action.
3. Exploit. Send the crafted prompt / poisoned content to the in-app assistant and confirm the model obeys it - leaked system prompt/secret, an unconfirmed tool call, or attacker-controlled output.
4. Observe the evidence. Trigger the vulnerable action and read the evidence panel - it prints the concrete proof (leaked value, accepted replay, executed payload, or unauthorized result).
5. Contrast with the secure path. Run the module’s secure/hardened action (where provided) and confirm the same attack is rejected - this is what a correct implementation should do.

Tools (optional)

• garak
• promptfoo

garak / promptfoo / MCP scanner test the LLM or MCP endpoint behind the app, not the app binary. Point them at the backend model API (find it with mitmproxy / Burp Suite) or a local on-device model server; the in-app demo already exercises the same prompt path.

Attack inputs

Payloads/artifacts you author for the on-device attack. The demo already ships and simulates these in-process (e.g. the malicious companion app / crafted intent is emulated inside the screen), so you only need to craft them to reproduce the exploit on a real device:

• crafted QR
• crafted text
• unicode tooling